GenAI gets adopted when it is baked into systems people already use
Lately I’ve been thinking about why some professions are rapidly getting value from GenAI while others – most? – have yet to see meaningful change in how they work. I believe the root cause is that GenAI is truly valuable when it is baked into systems those professionals already use. Law and medicine are two conservative professions that are both moving rapidly to GenAI rapidly. I’ve observed four structural factors pushing them along:
1.?????? Professionals in these industries already use software from a small number of well-resourced firms in their workflow. They can try out GenAI features without learning new software.
2.?????? The incumbent software makers have large, curated data stores that can’t be easily matched.
3.?????? These are the unusual professionals who bill by the 1/6th of an hour (lawyers) or patient procedures (doctors) so it’s easy to compute the ROI of efficiency gains.
4.?????? It’s illegal to practice law or medicine without a license, so the software companies can’t replace the humans; they have to support them.
These four factors create a happy confluence that is driving rapid adoption of GenAI. They aren't present in most other industries which will be the subject of next week's newsletter.
GenAI and the law: Law firms are highly fragmented (450,000 US law firms averaging only 3.3 lawyers per firm) but there are two huge legal research services: WestLaw from ThomsonReuters and LexisNexis. Over the years, they have invested building curated databases of statutes and rulings. They have rapidly developed GenAI upgrades to synthesize case law, draft briefs, and compare contracts. Lawyers working on large civil cases use a handful of e-discovery systems to process the documents they exchange with the court and each other. These are using GenAI for summarization and improved search.
JDJournal reports that in early 2025, 32% of law firms?and?20% of in-house teams?are deploying AI, but overwhelmingly they are using specialized legal systems. According to a new analysis from Anthropic, the maker of Claude models, legal professionals account for less than 1% of use. Lawyers are using AI but overwhelmingly they are using it inside the software they already use.
The tasks being done by GenAI were primarily being done by early-career lawyers working under senior colleagues. A survey by ThomsonReuters and Georgetown Law shows that big firms are reducing their hiring of entry-level lawyers and instead hiring experienced lawyers who can do higher value work still out of reach of the AI.
GenAI and medicine: Healthcare is a notoriously hard industry to change so I find its rapid adoption of GenAI stunning. There are 330,000 physician group practices in the US with only 2.3 physicians each. For 15 years the federal government has used a combination of carrots and sticks to get doctors and hospitals to use electronic health records. Now, there are a handful of software makers who are deeply integrated into the entire healthcare ecosystem. The biggest is Epic Systems with data on 250 million patients.
Epic has rapidly incorporated GenAI into the clinical workflow. Epic’s GenAI drafts doctors’ replies to patient emails. It transcribes doctor-patient conversations and uses the transcript to suggest entries to the patient record. The net is that the doctor can spend the appointment looking at the patient vs. looking down at a laptop. The overarching goal is to reduce physician burnout by letting doctors focus on patients and having AI handle the mountain of paperwork that comes with every patient.
There are structural factors driving medicine and the law to adopt GenAI rapidly, but most professions aren’t like this. In my next post I’ll use the market research industry as a counterexample and speculate about how things will play out. I also have a plea for VCs to reconsider the siren call of “Go Big” and fund startups supporting vs. disrupting incumbents. ??